You are on page 1of 6

Mobile Application for the Recommendation of Furniture and

Appliances Through Augmented Reality to Improve The User


Experience in The Online Shopping Process
Ricardo A. Guevara Jordy J. Rojas David G. Quevedo
School of Systems and Computer School of Systems and Computer School of Systems and Computer
Engineering, Peruvian University of Engineering, Peruvian University of Engineering, Peruvian University of
Applied Sciences, Peru Applied Sciences, Peru Applied Sciences, Peru
u201618078@upc.edu.pe u201512406@upc.edu.pe pcsidque@upc.edu.pe

ABSTRACT 2022), June 15–17, 2022, Madrid, Spain. ACM, New York, NY, USA, 6 pages.
Due to the increase in online shopping, companies are looking https://doi.org/10.1145/3545897.3545898
to reinvent themselves and improve their ecommerce platforms.
However, some users still do not dare to make an online purchase 1 INTRODUCTION
mainly because they cannot physically view the product. As a result In recent years, the constant advance of technology has marked
of this, innovative technologies have started to emerge. Therefore, a change in the way humans perform their daily tasks and what
this project seeks to improve the customer’s user experience, using tools are used to accomplish them. A clear example of this is the
augmented reality and colour recognition to recommend products way products are currently bought and sold [1]. Despite the signifi-
in a mobile application. For the development of the application, cant increase in online shoppers, not all product categories have
ARCore was used as a software development kit (SDK) to display increased their sales. During social isolation, the products that have
products and take measures with augmented reality. In addition, been most in demand are health products, followed by technologi-
a colour recommendation was implemented using the Palette API cal products [2]. This is because there is still great insecurity on the
(Application Programming Interface) for colour recognition and part of the buyer. According to Mastercard, of the people who have
an algorithm based on colour theory to generate the combinations. not yet bought virtually, 31% do not dare to do so because they do
The usability of the application was evaluated by a group of users, not see the product in person, 29% simply distrust, 15% think that
who completed a form that was based on Nielsen’s heuristic prin- their card data will be stolen, 15% consider that shipping costs are
ciples. The main conclusions of the analysis of the responses are high and 13% believe that the order will not arrive [3]. Based on
that they would make purchases using the application without hav- this problem, the use of augmented reality (AR) is proposed, a tech-
ing to physically see the product, users understood the purpose of nology that has not yet been fully exploited, so that users have a
using augmented reality and colour recognition to obtain recom- tool that allows them to find the product they need, through recom-
mendations. In addition, they were able to use all the functionalities mendations, and can really visualize what the products would look
implemented in the application without the need for a user man- like, without leaving home. One of the limitations is compatibility
ual, which highlights the minimalist design and ease of use of the with the devices since, in addition to the fact that the solution is
application. only compatible with the Android operating system, augmented
reality is only available for devices that have version 7.0 or later.
CCS CONCEPTS This problem has not been solved because there is no application
• Applied computing → Electronic commerce; Online shopping. that satisfies the need to acquire products in a remote way that
offers a different user experience with high security and reliability.
KEYWORDS In this paper a solution is proposed that will allow users to view
E-business, E-commerce, online shopping, augmented reality products through augmented reality, obtain recommendations by
size, measuring the available space, also with augmented reality
ACM Reference Format:
and obtain recommendations by colour, generating combinations
Ricardo A. Guevara, Jordy J. Rojas, and David G. Quevedo. 2022. Mobile
Application for the Recommendation of Furniture and Appliances Through
based on colours recognized from a photograph. The application
Augmented Reality to Improve The User Experience in The Online Shopping was developed for Android devices with Google’s ARCore software
Process. In 2022 3rd International Conference on Internet and E-Business (ICIEB development kit (SDK) and Palette API in the Kotlin programming
language. The application is developed for Android devices with
Permission to make digital or hard copies of all or part of this work for personal or version 7.0 (Nougat) onwards that support the Google Play service
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full citation
for AR. In addition, Firebase services are used such as: Cloud Fire-
on the first page. Copyrights for components of this work owned by others than ACM store, for the non-relational database; Storage, for the storage of
must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, 3D models and images of the products and Authentication, for the
to post on servers or to redistribute to lists, requires prior specific permission and/or a
fee. Request permissions from permissions@acm.org. login management.
ICIEB 2022, June 15–17, 2022, Madrid, Spain Our contributions are the following:
© 2022 Association for Computing Machinery.
ACM ISBN 978-1-4503-9732-2/22/06. . . $15.00 • 1. Being able to visualize the products to buy with augmented
https://doi.org/10.1145/3545897.3545898 reality so that the user has a better perspective of the product.

1
ICIEB 2022, June 15–17, 2022, Madrid, Spain Ricardo Alfredo Aparicio et al.

• 2. To be able to recommend a product based on the colours of • 3. User control and freedom
an environment that will be obtained through a photograph • 4. Consistency and standards
taken by the camera of the mobile device. This is in order to • 5. Error prevention
recommend a product that combines and is in harmony with • 6. Recognition rather than recall
the environment in which you want to place the product. • 7. Flexibility and efficiency of use
• 3. To be able to recommend a product based on dimensions • 8. Aesthetic and minimalist design
such as height, width and length. Which will be measured • 9. Help users recognize, diagnose, and recover from errors
from the same application. This is in order to recommend • 10. Help and documentation
products that fit in the space that the user has available.
3 MAIN CONTRIBUTION
2 CONTEXT As mentioned above, ARCore was used to offer the user the ability to
2.1 Augmented Reality [4] measure the different spaces that they have available when wanting
to purchase a product and based on this, recommend a product that
According to Nowacki and Woda, AR combines the real world with
fits the previously measured space. In addition, the user can choose
the virtual world, and can be captured through a lens. For example,
to obtain a recommendation by colour, which consists of taking
this technology is used in the solution to allow the user to view the
a photograph of the environment he wants to place the product
products from their mobile device.
in, to carry out this functionality it was considered to recognize
the main colours of the image, combinations are generated and,
2.2 ARCore [5]
finally, matching products recommended. Finally, the user can view
It is Google’s platform for the development of applications with Aug- the products with AR to know how the product will look after it
mented Reality, it allows the device to recognize the environment is purchased. For the visualization of the products, two functions
and, based on that information gathering, create new experiences were generated for the loading and positioning of the product with
interacting with the user. ARCore has three important characteris- AR, loading the 3D models from Firebase Storage. Please see Fig. 1
tics: and 2.
• Motion tracking. On the other hand, for the colour recommendation functionality,
• Environmental understanding. first the user is allowed to open the device’s camera application and
• Light estimation. take a picture, from this image, 6 main colours are obtained, defined
ARCore can detect normal, smooth or vertical surfaces, like by the Palette API as: VibrantColor, DarkVibrantColor, LightVi-
tables or dividers, and makes them accessible to applications such brantColor, MutedColor, LightMutedColor, and DarkMutedColor.
as plans. ARCore can also decide the limits of each plane and make Once recognized, colour combinations such as: Complementary,
this data accessible to the application. For example, it is used in Split Complementary, Triadic and Analogous are generated (See
the solution for taking measurements of the available space using Fig. 2). Based on these results, all products that combine with the
environmental understanding. environment are recommended. Finally, for the recommendation
by size, ARCore is used since it allows obtaining the location of a
2.3 Palette API [6] point within a surface. To measure the length and width, the user is
It is a compatibility library that allows extracting the highlighted allowed to locate 2 points within the surface (See Fig. 3), from the
colours of an image, to make the application more attractive. It position of these points, the distance between them is calculated
can detect 6 main colours in an image and is generally used to and finally the measurement obtained is changed to centimetres
change the colours of the application’s design components such (See Fig. 4). To perform the height measurement, the user is allowed
as headings or subtitles. For example, it is used in the solution to locate a point on the surface, once located, he is allowed to man-
to recognize the colours of a photograph taken by the user, and ually increase or decrease the height until the desired measurement
based on that, generate combinations. From the aforementioned is found. It is important to note that the measurements obtained are
definitions, it is understood that it is possible to provide the user not exact, they have a margin of error of ± 2cm and, on the other
with a better shopping experience using technologies such as AR hand, when loading 3D models from Firebase, there is a delay of 2.5
and colour recognition, which not only help to visualise the product seconds when positioning the model to be viewed with augmented
before buying it but also help the user to decide what product is the reality.
one that can best be adapted to the space that they have available.
4 RELATED WORKS
2.4 Nielsen’s heuristic principles [7] 4.1 The Use of Marker-Based Augmented
Heuristics were created in 1990 by Jakob Nielsen. These 10 "laws" Reality in Space Measurement [8]
guarantee that all tasks and operations carried out in a software are The authors seek to find an algorithm to perform measurements
executed in an efficient and straightforward way, thus creating a using AR with markers. The application was developed with Unity
more intuitive and user-friendly interface. The heuristic principles and the OpenCV library. It begins with the functionality of recog-
are the following: nizing the markers with the device’s camera, once this is done, the
• 1. Visibility of system status system calculates the angle of the walls and the distance between
• 2. Match between system and the real world them. If the points are perpendicular to each other, another 2 points

2
Mobile Application for the Recommendation of Furniture and Appliances Through Augmented Reality to Improve The User
Experience in The Online Shopping Process ICIEB 2022, June 15–17, 2022, Madrid, Spain

Figure 1: Model display functions

must be captured, and they must be combined with the points al- 4.3 Iterative prototyping of urban cobuilder:
ready obtained previously to obtain the distance between the walls. Tracking methods and user interface of an
The authors obtained high precision measurements. Compared to outdoor mobile augmented reality tool for
our solution, ARCore allows us to dispense with the use of book-
marks, which is optimal for applications where usability and user
co-designing [10]
experience must prevail, since only the mobile device is needed to The authors developed an application for phones where the user
start using the application. can work collaboratively in the construction of urban designs us-
ing augmented reality with markers. In the first simulation, they
worked with photographs of buildings located in strategic places.
Each building represented a marker, however this proved insuffi-
cient due to the general instability of recognition of a photographed
building. Later, functions such as GPS, compass and gyroscope were
used for motion detection and alignment. It was finally determined
that the distance between the user and the building posed a prob-
lem for marker detection. Our solution, by using augmented reality
without markers, gives the user the possibility of being free to use
the application in any environment and space, so that the distance
4.2 Using Augmented Reality to Measure
does not mean a problem for its correct use.
Vertical Surfaces [9]
In this paper, the authors seek to obtain vertical surface measure- 4.4 AR Furniture: Integrating Augmented
ments using ARKit and Unity. The application allows the user to Reality Technology to Enhance Interior
locate 3 points on the plane, from this, lines are drawn between the
points generating a polygon, and the Shoelace formula is used to
Design using Marker and Markerless
calculate the area. The authors obtained results with an imprecision tracking [11]
of 6.2%. Compared to our solution, we obtained a margin of error In this paper, the authors seek to visualize a product in a real
of ± 2cm and the height measurement was implemented, which is environment from their homes using AR with and without markers.
necessary for the user to be able to fully measure the space they The developed application consists of 4 stages which are: The input
have available. of the video from the camera to send them to the other processes,

3
ICIEB 2022, June 15–17, 2022, Madrid, Spain Ricardo Alfredo Aparicio et al.

Figure 3: Recognition and generation of colour combinations

5 EXPERIMENTS
This project was developed using the agile Scrum framework, which
consisted of 9 Sprints of 3 weeks in each one. In each Sprint, func-
tional tests were performed on each user story and the documenta-
tion of the main artifacts. The result was a certificate from ITService,
Figure 2: Viewing 3D Models an internal company of the university that provides support and
support with certifications and software quality tests.

5.1 Experimental protocol


The user satisfaction tests were carried out with a survey that was
based on the 10 heuristic principles of Nilsen. The surveyed users
had Android devices that supported augmented reality. The first
thing that was done was to install the APK of the application. Once
the detection of the markers, the modelling of the positioning of the correct installation was validated, the main characteristics that
the objects where the image frame can be superimposed and the our proposal offers were indicated so that the user can put it to
interaction interface. As results it was obtained that when markers use on their device and then can answer the survey. In addition to
are used, the distance between them and the camera should not be these heuristic principles, which are based on usability and user
greater than 60 cm. On the other hand, when it comes to the solution experience, users were asked if they would be willing to use the ap-
without markers, it works better on light surfaces and the lighting plication’s functionalities to purchase a product without physically
should be between 100 and 300 lux. Compared to our solution, seeing it.
ARCore automatically regulates lighting and can detect several
types of surfaces, which allows the user to use the application 5.2 Theoretical evaluation
wherever they want without the need to search for an environment The interview experiment was conducted with 25 participants who
with specific lighting. had augmented reality compatible devices and had Android 7.0

4
Mobile Application for the Recommendation of Furniture and Appliances Through Augmented Reality to Improve The User
Experience in The Online Shopping Process ICIEB 2022, June 15–17, 2022, Madrid, Spain

Figure 4: Calculate distance between 2 points

Figure 5: Distance calculated between 2 points

5
ICIEB 2022, June 15–17, 2022, Madrid, Spain Ricardo Alfredo Aparicio et al.

or onwards. Once users fully interacted with the application and purpose of using AR. On the other hand, the margin of error in the
tested all the implemented functionalities, they were asked to fill measurement and the delay in visualising the objects with AR did
out a form. From the data obtained, it was concluded that 98% of the not mean discomfort for the user. Finally, all users agreed that they
users indicated that it was possible to use the application without would purchase products using the implemented functionalities.
the need for a manual or instructions. On the other hand, users As a future plan to improve the application, we could test other
indicate that they find the design of the application attractive and technologies that are compatible with more devices, our solution is
that they can use it by applying the previous knowledge acquired limited to certain versions of Android, which could result in other
in other shopping applications. Users also note that the app was users not being able to use it. On the other hand, we could imple-
able to recover from surface recognition errors and take photos ment the functionality to visualise and position several products at
for colour recommendation. Finally, all users indicated that the the same time, so that users can visualise a complete environment
measurement was accurate, this means that the margin of error and make the purchase of several products at the same time.
was not perceived by the users, which made a pleasant impression
on them. REFERENCES
[1] Forbes. 2018. New Technologies That Will Change How Con-
sumers Shop In Store. (January 2018). Retrieved May 10, 2021 from
5.3 Evaluation of the experimental comparison https://www.forbes.com/sites/michelleevans1/2018/01/18/new-technologies-
When analysing the responses of our interviewees for each heuris- that-will-change-how-consumers-shop-in-store/?sh=311c8f735fc9
[2] ECommerceNews. 2020. Comercio electrónico: El 47% de peruanos realiza sus
tic, we reached the following conclusions: compras de manera online en la nueva normalidad. (August 2020). Retrieved
May 10, 2021 from https://www.ecommercenews.pe/comercioelectronico/2020/
• 1. The technologies used for the development of the project peruanos-compras-online.html
allow to meet the quality attributes of availability, perfor- [3] Mastercard. 2020. Electronic Commerce and online services as an alter-
mance and usability. native for a society in the process of digitization. Retrieved May 11, 2021
from https://www.mastercard.com/news/latin-america/es/perspectivas/blog-
• 2. Users understand the purpose of using augmented reality posts/blog-es/2020/marzo/el-comercio-electronico-y-los-servicios-en-linea-
for taking measurements and displaying products. In addi- como-alternativa-para-una-sociedad-en-proceso-de-digitalizacion/
tion, they deduce how to use the application without the [4] Intel. Virtual Reality vs. Augmented Reality vs. Mixed Reality. Retrieved May
10, 2021 from https://www.intel.com/content/www/us/en/tech-tips-and-tricks/
need for a manual or guide. virtual-reality-vs-augmented-reality.html
• 3. Users find the design and flow of the application attrac- [5] ARCore. 2021. Overview of ARCore and supported development environments
(November 2021). Retrieved May 10, 2021 from https://developers.google.com/ar/
tive, however, some interviewees indicate that it could be discover
improved. [6] Developers. 2020. Selecting Colors with the Palette API. (January 2020). Re-
• 4. The margin of error in taking measurements did not mean trieved May 10, 2021 from https://developer.android.com/training/material/
palette-colors
an inconvenience for users when interacting with the appli- [7] NN/g Nielsen Norman Group. 2020. 10 Usability Heuristics for User Interface
cation. Design. (November 2020). Retrieved May 10, 2021 from https://www.nngroup.
• 5. The small delay when loading the products to be viewed com/articles/ten-usability-heuristics/
[8] Salin Boonbrahm, Poonpong Boonbrahm, and Charlee Kaewrat. 2020. The Use
with augmented reality was not highlighted as a defect in of Marker-Based Augmented Reality in Space Measurement, Procedia Manufac-
the interviews. Users are willing to purchase a product using turing. 337-343. https://doi.org/10.1016/j.promfg.2020.02.081.
[9] Bergquist Robin and Stenbeck Nicholas. 2018. Using Augmented Reality to Mea-
the functionalities of our solution, without having to have sure Vertical Surfaces (Dissertation). Retrieved from http://urn.kb.se/resolve?
previously seen the product. urn=urn:nbn:se:liu:diva-148868
[10] Hyekyung Imottesjo, Liane Thuvander, Monica Billger, Peter Wallberg, Gustav
Bodell, Jaan-Henrik Kain, and Stig Anton Nielsen. 2020. Iterative Prototyping
6 CONCLUSIONS AND PERSPECTIVES of Urban CoBuilder: Tracking Methods and User Interface of an Outdoor Mo-
bile Augmented Reality Tool for Co-Designing. Multimodal Technologies and
The project consisted of the development of a mobile application Interaction 4, 2: 26. https://doi.org/10.3390/mti4020026
with augmented reality for Android devices, which allows users to [11] Waraporn Viyanon, Thanadon Songsuittipong, Phattarika Piyapaisarn, and Suwa-
have a better online shopping experience. With the application, the nun Sudchid. 2017. AR Furniture: Integrating Augmented Reality Technology
to Enhance Interior Design using Marker and Markerless tracking. In Proceed-
user will obtain product recommendations according to their needs, ings of the 2nd International Conference on Intelligent Information Processing
such as a recommendation by colour of a room or a recommenda- (IIP’17). Association for Computing Machinery, New York, NY, USA, Article 32,
tion by size in a specific space. Once the user has found the product 1–7. https://doi.org/10.1145/3144789.3144825

that suits their needs, the application allows them to visualise it


with augmented reality. Google’s ARCore SDK was used for this,
as this is the tool that recognises the largest surfaces. Compared
to other applications, our solution gives the user the freedom to
use the application at any time and in any space, as it makes use of
AR without markers. Moreover, it can not only be used to preview
products, but also serves as an automatic sales guide or advisor that
helps the user to find the right product according to their needs.
To carry out the experiment, surveys were conducted with users
who had already installed and interacted with our solution. As a
result of these surveys, it was found that users could use the appli-
cation without impediments and without the need to read a user
manual or instruction manual. Furthermore, they understand the

You might also like